Use of fluorescence in situ hybridization to determine the relationship between chromosome aberrations and cell survival in eight human fibroblast strains

A predictive assay of normal tissue radiosensitivity could benefit 'treatment tailoring' of radiotherapy for certain categories of tumour. The use of present clonogenic cell survival assays for this purpose would be impractical in routine clinical practice because of the lengthy assay time. Fluorescence in situ hybridization (FISH) using whole chromosome probes on metaphases was investigated as a potential substitute. Eight human fibroblast cell strains with a range of radiosensitivities were tested. For each strain, cell survival curves were determined and correlated with chromosome aberrations detected by FISH performed on metaphase cells collected 52 h after irradiation. A whole chromosome probe for chromosome 4 was used for all cell strains. The results revealed an increase in the percentage of metaphases with aberrant chromosomes (translocations and/or breaks) with increasing radiation dose for all strains. For the more radiosensitive cell strains there were relatively more aberrant metaphases for a given radiation dose when compared with fibroblasts from a normal donor. The relationship between surviving fraction and chromosome aberrations showed some variation between strains, but a linear regression for all data showed a highly statistically significant correlation (r = 0.89, p < 0.0005). These results suggest that an assay of chromosome damage using FISH could substitute for the clonogenic assay to predict the radiation sensitivity of human fibroblasts.

Defective repair of DNA double-strand breaks and chromosome damage in fibroblasts from a radiosensitive leukemia patient

A radiation-sensitive fibroblast culture (180BR) established from an acute lymphoblastic leukemia patient who died following radiotherapy is defective in the repair of radiation-induced DNA double-strand breaks. The cells also show a reduced capacity to repair interphase chromosome damage visualized by means of premature chromosome condensation and metaphase chromosome aberrations measured by fluorescence in situ hybridization on chromosome 4. This case represents the first example in humans where hypersensitivity to ionizing radiation can be ascribed directly to a defect in DNA and chromosome repair, and the defect may underlie the cancerous phenotype observed.

Can measurements of potential doubling time (Tpot) be compared between laboratories? A quality control study

The purpose of this study was to investigate the reproducibility of potential doubling time measurements of human tumors in different laboratories and to distinguish which steps in the measurement procedure were subject to the greatest variation. This was achieved by comparing measurements on the same source material in two different laboratories in which three aspects of the technique were separately studied, namely, preparation and staining of the nuclear suspensions, running the samples on the flow cytometer (FCM), and analyzing the two-parameter FCM data. This involved exchange between the two centers of fixed tumor material, stained nuclear suspensions, and FCM data on floppy disks. The analysis step was found to be the least variable step. For DNA synthesis time, Ts, and the labeling index, LI, the coefficients of determination (R2) ranged from 92% to 95.4%. A small systematic bias was observed, with one center measuring approximately 15% higher values for both LI and Ts. Different criteria for window placements were found to be a contributing factor. Variations in the FCM step were approximately equal to those for analysis, with no systematic deviations. Variations for the preparation and staining step were the largest (R2 = 60.5% and 38.1% for LI and Ts, respectively). However, this step was the only one subject to intratumoral variability, which was the largest contributing factor to the variations observed. In addition, however, LI was on average 41% higher in one center compared to the other, resulting in a systematic bias. Differences in the level of green fluorescence of the labeled cells implicated antibody differences as a possible cause. The variations found here for the three procedural aspects were significantly smaller than variations observed between tumors, a requirement for a predictive assay. Suggestions for implementation of quality control procedures include objective (computer-assisted) data analysis on two-parameter histograms and optimization of antibody combinations.

The effect of radiation on G2 blocks, cyclin B expression and cdc2 expression in human squamous carcinoma cell lines with different radiosensitivities

The purpose of the present study was to investigate the role of cyclin B and cdc2 in the G2 delay and to test whether the magnitude of the G2 delay correlated with sensitivity to ionizing radiation in two human cell lines. Cell cycle delays were measured by flow cytometry after pulse labeling with bromodeoxyuridine, and expression of cell cycle control genes were measured in Western blots in radiosensitive SCC61 and radioresistant SQ20B cell lines. Flow cytometry data demonstrated that the duration of the G2 arrest was dose dependent in both cell lines, amounting to approximately 1.1 h/Gy. No difference was found between the cell lines in the length of the G2 block. Radiation exposure did not result in a decrease of cyclin B. Cyclin B protein levels in both asynchronous and synchronized populations in fact showed a dose dependent increase, concomitant with the rise in the fraction of cells in G2/M. Similarly, the cdc2 protein levels did not decrease after irradiation. However, it was found that the levels of hyperphosphorylated, and therefore inactive, kinase were significantly higher in irradiated cells than in unirradiated cells. The accumulation of this hyperphosphorylated form correlated with the arrest of cells in the G2 phase. Finally, immunocytochemical staining of cyclin B revealed an increase of this protein in the cytoplasm after irradiation and a decrease in nuclear staining. This differential localization could possibly account for the reduced nuclear phosphorylation of cdc2 kinase leading to the G2 arrest.

Differential repair of radiation-induced DNA damage in cells of human squamous cell carcinoma and the effect of caffeine and cysteamine on induction and repair of DNA double-strand breaks

The goal of these experiments was to investigate further the relationship between DNA double-strand breaks and cell killing in human tumor cells, first by comparing different cell lines, and second by radiomodification studies. Field-inversion gel electrophoresis was used to quantify double-strand breaks. Two subclones of the radioresistant human squamous cell carcinoma line SQ20B (SQD9 and SQG6) were compared. These subclones differed in DNA index by a factor of 1.7 but showed the same resistance to radiation as cells of the parental cell line. It was found that, although induction of DSBs was not significantly different in the two cell lines, the t1/2 of the fast component of repair was significantly shorter for SQD9 cells, leading to greater overall repair which was not reflected in increased survival. Caffeine and cysteamine were tested as modifiers of radiosensitivity, using the radioresistant SQ20B line and the radiosensitive SCC61 cell line. No effect of caffeine was seen when the drug was present only during irradiation. Postirradiation incubations with caffeine, however, resulted in a dose reduction factor greater than 2.0 in cell survival for both cell lines. In contrast, induction of DSBs was reduced by caffeine, and no effect on DSB repair was observed. Cysteamine led to a dose protection factor greater than 1.8 in cell survival in both cell lines. A reduction in induced DSBs was found at high doses corresponding approximately with the increase in cell survival. Over the same (low) dose range, however, the correlation between DSB induction and cell killing was poor. These data indicate that DSB induction does not correlate well with cell killing either for different cell lines, for radiochemical modification (cysteamine) or for some other types of modification (caffeine).

Use of fluorescence in situ hybridization to measure chromosome aberrations as a predictor of radiosensitivity in human tumour cells

Fluorescence in situ hybridization (FISH) is a potential assay for determining cellular radiosensitivity based on the detection of chromosome damage. This approach was chosen because of its relative simplicity and short assay time. Two radiosensitive and two radioresistant human tumour cell lines were used. The radiosensitive lines were an ovarian carcinoma line (A1847) and a squamous carcinoma line (SCC61). The radioresistant cells were a lung adenocarcinoma line (A549) and a second squamous line (SQ20B). Whole chromosome-specific probes were used to detect radiation-induced chromosome aberrations in mitotic cells. Available probes were first screened to characterize the intrinsic chromosome aberrations before irradiation and the appropriate probes (minimum fluorescent spots) were selected for each cell line. Maximum radiation-induced aberrations were found 24 h after irradiation. Dose-response curves corrected for target size (proportion of genome probed) differed for all cell lines. The radiosensitive A1847 cell line showed more induced aberrations compared with the radioresistant A549 cell line, in agreement with the survival data. In contrast, the SQ20B cell line showed more induced chromosome aberrations than the more radiosensitive SCC61 cell line, leading to the hypothesis that the SQ20B cells could tolerate more aberrations. Dose-response curves obtained in surviving cells 14 days postirradiation indeed showed elevated levels of chromosome aberrations for SQ20B cells. The difference in chromosome aberrations between 1 and 14 days showed a good correlation with the survival data for all four cell lines. In conclusion, FISH of mitotic cells with whole chromosome probes appears to be a suitable assay to predict radiosensitivity. It seems necessary, however, to determine both induced and remaining chromosome aberrations, since different processing or tolerance of radiation-induced aberrations, including stable types, could lead to different correlations with cell survival.

Improvement of combined modality therapy with cisplatin and radiation using intratumoral drug administration in murine tumors

The aim of these studies was to increase the therapeutic ratio by achieving higher tumor concentrations of cisplatin during the course of a fractionated irradiation treatment. Specific goals were to test, firstly, whether multiple drug injections could be replaced by a single slow release implant of cisplatin, and secondly, whether the therapeutic potential of the combined treatment could be increased by administering the drug intratumorally. Drug administration routes tested were intraperitoneal (i.p.) of drug in solution, intratumoral (i.t.) of drug in solution, and intratumoral of drug in a slow release formulation. The latter consisted of a hydrogel polymer formulated into rods which were implanted into the center of subcutaneous tumors. For drug alone, both i.t. routes (solution or polymer) produced higher therapeutic gains than i.p. administration, as judged by tumor growth delay for a given weight loss. When combined with radiation, dose response curves were always shifted to lower doses and were steeper than for radiation or drug alone, although isobologram analysis indicated additivity. In a first series, drug enhancement ratios ranged from 1.6 to 2.6, and were highest for the i.t. groups. In a second series, X-ray enhancement ratios ranged from 1.1 to 1.7, with overlap between results from the different routes. Therapeutic ratios, however, were highest for the i.t. groups in both series. Slow release rods produced the highest therapeutic gains in the first series, while i.t. administration of drug in solution was approximately as effective in the second series.(ABSTRACT TRUNCATED AT 250 WORDS)

Pharmacokinetics in melanoma-bearing mice of 5-dihydroxyboryl-6-propyl-2-thiouracil (BPTU), a candidate compound for boron neutron capture therapy

Blood pharmacokinetics and tissue distribution of 5-dihydroxyboryl-6-propyl-2-thiouracil (BPTU), a boron carrier with postulated melanin-seeking properties for boron neutron capture therapy, were determined in C57/BL mice with subcutaneous pigmented or non-pigmented B16 melanomas. Borocaptate sodium (BSH) was used as a boron compound without melanin-seeking properties in a comparative biodistribution study in the same animal tumour models. Administration of single doses showed that BPTU was retained better in the pigmented B16 tumour than in the non-pigmented variant. BPTU was found in large concentrations in kidney and liver. Brain boron was approximately 10-fold lower than tumour boron. On a molar basis, BPTU demonstrated higher affinity for B16 tumours than BSH. Owing to solubility limits, tumour boron concentrations in this mouse study were too low for effective application of BNCT. However, the high tumour-to-blood and tumour-to-normal tissues ratios indicate that, with appropriate formulation, BPTU could be a promising candidate for clinical BNCT.

Intrinsic radiosensitivity and chromosome aberration analysis using fluorescence in situ hybridization in cells of two human tumor cell lines

The survival curves for cells of two human tumor cell lines, HT29 and MeWo, have been defined using a Dynamic Microscopic Imaging Processing Scanner (DMIPS). There are two major differences between these two cell lines: (a) HT29 is more radioresistant than MeWo (surviving fraction at 2 Gy of 74 and 27%, respectively) and (b) HT29 presents a marked multiphasic survival curve with hypersensitivity at low doses (< 0.5 Gy) followed by an increase in radioresistance at higher doses which we have interpreted as "induced radioresistance"; this phenomenon is much less pronounced for the more radiosensitive cell line MeWo. We have now measured in these two cell lines the stable chromosomal aberrations and fragments, with the method of fluorescence in situ hybridization (FISH). We have analyzed chromosome 4, which does not have spontaneous translocations in either of these two cell lines. A dose-effect relationship was studied for radiation doses up to 5 Gy. At all doses, both translocations and breaks are more frequent in the radiosensitive cell line MeWo compared to the radioresistant cell line HT29. The correlation between survival and translocations is different for HT29 and MeWo, thus indicating that another factor(s) may be involved in cell killing in these lines.

Quantification of patient to patient variation of skin erythema developing as a response to radiotherapy

A method is described to determine accurately skin redness during a course of radiotherapy using reflectance spectroscopy utilizing information from across the visible spectrum according to the L*a*b* color coordinate system. The method was used to quantify the development of skin erythema during and after electron beam irradiation of the chest wall following mastectomy. A number of factors were identified which could influence the wide variation in response seen between patients. These were: intra- and inter-observer variation; intra- and inter-patient variation and variation in the actual dose delivered. Statistical analysis, including an analysis of variance of inter- and intra-patient variation, revealed that the major factor that accounts for the observed difference between patients is a true inter-patient variation, with a coefficient of variation, corrected for intra-patient variation, of 43%. Within the narrow dose range administered in this study, there was no demonstrable dose-effect relationship, raising questions about the role of cell death in the basal layer of the epidermis in the pathogenesis of radiation induced erythema.

A comprehensive PC-based computer model for microdosimetry of BNCT

A computer model is described that performs microdosimetric calculations of the radiation dose delivered to tumour and normal tissue in boron neutron capture therapy (BNCT) by simulating capture reactions in a predefined three-dimensional space. The role of intracellular boron distributions and cellular dimensions on the radiation dose in clinical and experimental BNCT has been studied using a PC-based computer model. In order to calculate the radiation dose to low boron uptake cells, the extent of irradiation by boron containing adjacent cells (cross fire) is also dealt with. Radiation doses from boron and nitrogen neutron capture are converted to a biological effect by means of relative individual ion track segment efficacies, based on linear energy transfer along the particle track. A good correlation was found after comparing predicted values with previously published experimental data. A number of examples is given to illustrate the program's features.

Characterization of resistance mechanisms to cis-diamminedichloroplatinum(II) in three sublines of the CC531 colon adenocarcinoma cell line in vitro

Cisplatin resistance was developed in sublines of the CC531 rat colon adenocarcinoma cell line by continued low level drug exposure. Two relatively stable lines were obtained (RL2 and RL4) which were 6- and 20-fold more resistant to cisplatin. In addition, a subline more sensitive than the parent line by a factor of 2 (RLS) was obtained by subculture from a treated tumor. Mechanisms of resistance to cisplatin were investigated in these four lines, with the aim of determining the relative contributions of different resistance mechanisms at various resistance levels. Drug accumulation linearly decreased with increasing drug resistance. A 20-fold resistance was associated with only a 5-fold decrease in accumulation, suggesting that other resistance mechanisms may be involved in the total degree of resistance. Intracellular glutathione, measured fluorometrically, also increased with increasing resistance, varying by a factor of 4 between the most and least resistant lines. Reduction of glutathione levels by buthionine sulfoximine to parent line levels increased sensitivity but the cells remained considerably more resistant than parent cells. Resistant lines cultured in the absence of drug became progressively more sensitive, without accompanying changes in total glutathione levels. DNA-drug adducts, the presumed toxic lesion, were measured immunocytochemically. Initial levels decreased with increasing platinum resistance, although not proportional to resistance (factor of 5 decrease for 20-fold resistance). Drug dose ratios for equal initial adducts were similar to dose ratios for equal drug accumulation, implying that intracellular concentrations solely determine DNA adduction and that differences in glutathione level had little influence on the proportion of drug which eventually formed adducts. After 48 h, a better correlation between remaining adducts and resistance was found (factor 12 less adducts for 20-fold resistance). This implies that repair of adducts was important in determining survival. These data indicate that decreased drug accumulation played a proportionally greater role in the moderately resistant cell line and that adduct repair played a progressively greater role in the highly resistant cell line.

Prediction of repopulation rates and radiosensitivity in human tumours

An accurate prediction of response to radiotherapy would allow treatments to be modified for those individuals where a relatively poor response is expected with conventional treatment. Present assays receiving most attention are those for repopulation, radiosensitivity and hypoxia. This paper discusses our experience with the first two, together with relevant data from other groups. Attention is focussed particularly on correlations with outcome in clinical trials, and the accuracy of present assay methods.

Cell kinetic analysis of murine squamous cell carcinomas: a comparison of single versus double labelling using flow cytometry and immunohistochemistry

The study was originally set up to measure accurate cell kinetic parameters in two murine squamous cell carcinomas (scc) for comparison with radiobiological data on proliferation during radiotherapy. The tumours, AT84 and AT478, were both moderately well differentiated aneuploid scc. In the course of the study, several comparisons of techniques were made in two different centres. This paper reports on the results of those comparisons involving two different detection methods (flow cytometry and immunohistochemistry), single vs double labelling, and in vivo and in vitro labelling, the latter using tissue slices incubated under high pressure oxygen. Pulse labelling studies with bromodeoxyuridine (BrdUrd) showed that the labelling indices (LI) were not significantly different after in vitro or in vivo labelling. In addition, the flow cytometry (FCM) and immunohistochemistry (IHC) methods also gave labelling indices which were not significantly different. Only tumour cells were analysed in these studies by selecting cells on the basis of aneuploidy (FCM) or morphology (IHC). The DNA synthesis time of the tumour cells were analysed by both techniques. For FCM, the Relative Movement method was used (Begg et al., 1985). For IHC, a double labelling method was used, employing BrdUrd and triated thymidine (3H-TdR) administered several hours apart, detected simultaneously using immunoperoxidase and autoradiography, respectively. When both labels were administered in vivo, there was good agreement for Ts between the FCM and IHC methods. Attempts were also made to measure Ts in vitro using both techniques. With double labelling, it was found that cells did not take up the second label, implying a failure of cycle progression. This was confirmed by FCM results, showing no movement of labelled cells through the S-phase, despite an initially high uptake. This could not be influenced by lowering the DNA precursor concentration or by adding foetal calf serum. This indicates that DNA synthesis times are difficult or impossible to measure in vitro in fresh tumour explants. Finally, the double labelling IHC method allowed intratumoural variations of both LI and Ts to be studied. Both parameters were found to vary markedly throughout the tumour volume, particularly for larger tumours (600 mg), giving calculated local potential doubling time values (Tpot) ranging from 1-7 days.

Lack of correlation of human fibroblast radiosensitivity in vitro with early skin reactions in patients undergoing radiotherapy

Fibroblasts from breast cancer patients were obtained as outgrowths in vitro from punch biopsies and their radiosensitivity tested in early passages. Skin erythema reactions in the same patients were also measured, as degree of redness using reflectance spectrophotometry. Measurements were taken before and during a 4-week radiotherapy treatment with electrons to the thoracic wall. Of 59 biopsies studied, radiosensitivity and erythema were concurrently studied in 32. In 24, evaluable data from both clinic and laboratory were obtained. A population growth assay in 96-well plates, using absorption of sulphur rhodamine B as the stain for cell numbers, showed good agreement with the colony-formation assay. Plating efficiencies and growth rates in the colony assay were higher using human serum in place of foetal calf serum. Cell survival curves with human serum were mostly exponential with little shoulder. The parameters of survival at 2 Gy (SF2) and the dose required to give 10% survival (D10) were used in the correlations with clinical data; these were 0.25 +/- 0.09 and 3.03 +/- 0.50 Gy, respectively. There was a strong correlation between these two survival curve parameters (r = 0.98). Skin redness was found to linearly increase with time during radiotherapy. The slope of the increase differed markedly from patient to patient, with a range of a factor approx. 10. No correlation was found between SF2 and erythema response in the 24 evaluable patients (r = 0.13, p > 0.5). A similar lack of correlation was found using D10 as the radiosensitivity parameter (r = 0.12, p > 0.5). These data indicate that fibroblast radiosensitivity measured in vitro cannot be used to predict erythema reactions to radiotherapy in breast cancer patients.

Radiation-induced DNA damage and repair in radiosensitive and radioresistant human tumour cells measured by field inversion gel electrophoresis

Radiation-induced DNA damage induction and repair was measured in two human squamous carcinoma cell lines with differing radiosensitive. Experiments were carried out with field inversion gel electrophoresis (FIGE), adapted to measure DNA double strand break (DSB) induction and repair in unlabelled cells. The sensitivity of the method was increased by introducing a hybridization membrane into the agarose gel. Damaged DNA accumulated on one spot on the membrane resulting in high local concentrations. This DNA was quantified using radioactively-labelled total human DNA as a probe. Dose response experiments for damage induction correlated well with the results using prelabelled cells. Linear DNA damage induction curves were observed with a sensitivity for the post-labelling method of 1 Gy. No differences in DSB induction were found, however, between the radiosensitive SCC61 and the radioresistant SQ20B cell line. Repair experiments were carried out with trypsinized cells with different doses and repair temperatures. The 10, 25 and 50 Gy doses resulted in 6, 13 and 50% of the DNA migrating out of the plug at 0 h. For both the cell lines 75-85% of the initial damage was repaired within 1 h at 37 degrees C at all three radiation doses, i.e. no significant differences were observed in repair rates or extent between the two cell lines. At 24 degrees C repair was slower than at 37 degrees C, and at 0 degree C no repair was observed. In summary, radiosensitivity differences at physiological temperatures could not be explained by differences in either induction or repair of DNA damage as measured by pulsed field gel electrophoresis.

Ploidy status and the response of T1 glottic carcinoma to radiotherapy

Flow cytometric DNA ploidy measurements were performed on formalin fixed paraffin embedded tumour specimens from patients with a T1 glottic laryngeal carcinoma in order to evaluate the role of DNA content in relation to local control. From 1980 to 1987, a consecutive series of 90 patients with a T1 glottic laryngeal carcinoma were treated by radiotherapy with curative intent. Biopsies from 44 of these patients were readily available for DNA flow cytometry. In this group aneuploidy was associated with a significantly higher risk (P = 0.018) of local recurrence within 2 years after completion of radiotherapy (38% vs. 9% in the diploid group).

Selectivity of boron carriers for boron neutron capture therapy: pharmacological studies with borocaptate sodium, L-boronophenylalanine and boric acid in murine tumors

Borocaptate sodium (BSH) and L-boronophenylalanine (L-BPA) are two boron carriers used for boron neutron capture therapy (BNCT) in the treatment of glioblastoma and melanoma, respectively. The suitability of these two compounds was evaluated on the basis of pharmacokinetic studies aiming at characterizing their biodistribution, tumor uptake and tumor selectivity. Boric acid was also used as a reference compound since it is nonselective and relatively freely diffusible. The compounds were investigated in two tumor models, a B16 pigmented melanoma and the RIF1 sarcoma. Mice were sacrificed after different boron doses at various post-injection times and tissue and plasma levels measured using inductively coupled plasma atomic emission spectroscopy (ICP-AES). The proposed minimum effective tumor boron concentration of 15 ppm was achieved in both tumor models for the three compounds tested, although only for L-BPA in the melanoma was this achieved when tumor-plasma ratios were above 1. In the RIF1 model, maximum tumor concentrations of 44 and 31 ppm B were reached after administration of 50 micrograms B/g body weight for boric acid and BSH, respectively. After administration of 12.5 micrograms B/g of L-BPA, maximum concentrations of 15 and 21 ppm were found in the RIF1 and B16 models, respectively. Tumor-plasma ratios (TPR) for BSH remained close to or below unity at all times studied in both tumors. Brain levels of BSH were very low, however, leading to tumor-brain ratios markedly greater than 1 at all times. L-BPA and boric acid showed TPR values above unity in both tumor models, reaching 3.2 in B16.(ABSTRACT TRUNCATED AT 250 WORDS)

Reduction of systemic exposure and toxicity of cisplatin by encapsulation in poly-lactide-co-glycolide

The tissue distribution and normal tissue toxicity of cisplatin (cDDP) administered as poly-lactide-co-glycolide (PLAGA) microspheres, developed for loco-regional administration of cDDP to the liver, were studied in Wag/Rij rats. Venoportal administration of this formulation resulted in a reduction in total systemic and renal toxicity, which correlated with a decrease in normal tissue exposure to cDDP while maintaining high liver platinum levels. Liver-to-kidney platinum level ratios were 28 times higher after 4 h and 19 times higher after 24 h with PLAGA-cDDP microspheres than with free cDDP. Liver-to-blood platinum ratios at these times were 38 times and 36 times higher using PLAGA-cDDP. In a CC531 colon carcinoma liver micrometastases model, cytotoxicity of microsphere-released cDDP was confirmed in vivo by equal inhibition of tumor growth by PLAGA-cDDP and free cDDP over a period of 26 days. Free cDDP, however, caused significantly more histological renal damage and total body weight loss. The results were supported by the finding of higher plasma creatinine and urea concentrations 26 days after administration of free cDDP. Kidney platinum levels were 7 times lower when PLAGA-cDDP was used. These findings indicate a sparing effect on normal tissues when cDDP is targeted to the liver by formulation in PLAGA. PLAGA-cDDP microspheres may, therefore, be a useful and effective addition to current techniques of loco-regional chemotherapy for disseminated hepatic tumors.

Intratumoural administration of cisplatin in slow-release devices: II. Pharmacokinetics and intratumoural distribution

The pharmacokinetics of cisplatin in mice with s.c. RIF 1 tumours was studied after intratumoural (i.t.) administration of drug in solution and in different slow-release devices. The data were compared with those obtained after i.p. administration of cisplatin. The slow-release devices under test were manufactured from either starch (ST) or polymeric hydrogels with different water uptakes (named T1, T2 and T3). In vitro release from these devices was approximately 100% in 2 h for starch rods, 100% in 24 h for T3 hydrogels, 45% in 4 days for T2 hydrogels and less than 10% in 4 days for T1 hydrogels. In vivo release rates agreed well with the in vitro data for T1 and T2 rods and were slightly slower in vivo for the T3 rods. The ST rods released the drug 6 times slower in vivo than in vitro. Plasma concentrations after i.t. administration were lower than those measured after i.p. administration. Systemic exposure to both total and free platinum was reduced to 70% for i.t. as compared with i.p. administration. Tumour concentrations were 4 times higher after i.t. than after i.p. administration. Tumour and peak plasma levels of platinum increased with increasing release rates. With the faster-releasing formulations (ST and T3), tumour platinum concentrations were 100 times higher than after i.p. administration. With the slower releasing formulations (T1 and T2), total tumour platinum concentrations were 2 and 9 times higher, respectively, than after i.t. administration of cisplatin in solution. Platinum distribution within the tumour was homogeneous after i.p. administration. After i.t. administration of cisplatin in solution, platinum concentrations in the centre of the tumour were approximately 4 times higher than in peripheral tumour tissue. Implantation of cisplatin in T2 and T3 hydrogel rods resulted in large concentrations of platinum in the centre of the tumour (the site of implant), which decreased steeply towards the tumour periphery. In summary, i.t. administration of cisplatin solution produced better results than did systemic (i.p.) administration in terms of tumour versus plasma drug-concentration ratios. Administration of drug in slow-release rods proved even more advantageous, although this was offset by inhomogeneous drug distributions within the tumour.